NTNU Norwegian University of Science and Technology Faculty of Humanities Department of Language and Literature
Master ’s thesis
Kine Barli Mjøsund
A Descriptive Comparison of the Swarm Alternation in English and Norwegian
Master’s thesis in Language Studies with Teacher Education Supervisor: Christopher Wilder
June 2020
Kine Barli Mjøsund
A Descriptive Comparison of the Swarm Alternation in English and Norwegian
Master’s thesis in Language Studies with Teacher Education Supervisor: Christopher Wilder
June 2020
Norwegian University of Science and Technology Faculty of Humanities
Department of Language and Literature
Preface
My interest in languages, though especially the English language, is something I discovered when I was fairly young, and my eagerness to learn more and more never stopped. Though, it was after I went to Australia as an exchange student when I was still in high school that I was absolutely certain; I wanted to study English at the university and become an English teacher. At that point, however, I was only imagining that further down the road I will be writing a master’s thesis focused on one of the topics that
interest me most.
The thesis was written largely within the time frame September 2019 to May 2020. I have not carried out any surveys or questionnaires, though the data largely comprise three types of sources: data presented in the relevant academic literature; attested natural data, found e.g. via Google search; and native speaker judgements on attested and constructed examples. As my thesis supervisor is a native speaker of English, he has provided judgements on the English data, whereas judgements on the Norwegian data have mostly been done by me, in addition to occasionally asking friends and family.
Furthermore, since no work, to my knowledge, has previously been done on the swarm alternation in Norwegian, the research and theories of other linguistic scholars (e.g.
Salkoff (1983); Levin (1993); Dowty (2000); Rowlands (2002)), are central as a point of departure for the investigation of similar aspects of the alternation in Norwegian.
Based on previous research on the swarm alternation in English, the main focus of the thesis has been to write a descriptive comparison aiming at investigating the nature of the phenomenon in Norwegian as well as discover possible English-Norwegian similarities and differences in this domain. Although the deeper questions that may arise are not easily answered within the scope of this thesis, I hope that I have been able to provide some substantial groundwork for further studies and research.
Some days have of course been easier than others when it comes to the process of writing this thesis, however, I am very grateful that I have had the opportunity to carry out a project which has been very meaningful to me, which in turn has provided me with even more knowledge about a topic I am very interested in.
Acknowledgement
Throughout the process of writing this master’s thesis I have spent numerous hours alone discovering some interesting aspects and making sense of my own thoughts in front of my computer, however, there are a few specific people who have been very important along the way.
The first person I would like to give special thanks to is my master’s thesis supervisor, professor Christopher Wilder, who is the person responsible for sparking my interest for studying the swam alternation by enlightening me about the world of linguistics, though especially Generative Grammar and Argument Structure alternations, through courses such as ENG2155 Theoretical Approaches to English Language and ENG3110 Special Subject in English Language. He has truly been invaluable to the entire process, giving me high quality guidance and supervision along the way, and his knowledge about relevant literature, theories and the world of linguistics in general has been a remarkable resource.
A project like this also requires access to the relevant literature, which is why I secondly would like to thank Dragvoll Library for providing a large amount of the literature
needed.
Last, but not least, I would like to thank my family and friends, though perhaps especially one of my best friends and roommate, for being very supportive along the way, as well as providing me with some native speaker judgements on the Norwegian data.
Thank you.
Kine Barli Mjøsund Trondheim, spring 2020.
Table of Contents
Preface ________________________________________________________________ v Acknowledgement _______________________________________________________ vii Table of Contents ________________________________________________________ ix List of Tables ____________________________________________________________ x 1. Introduction __________________________________________________________ 1 1.1 Topic and Approach _________________________________________________ 1 1.2 Main Points ________________________________________________________ 1 1.3 Structure __________________________________________________________ 2 2. Background___________________________________________________________ 3 2.1 Argument Alternations _______________________________________________ 3 2.2 Introduction of Basic Structure: The Syntactic Pattern(s) in Question __________ 4 2.3 The Lexicalist Approach and the Role of Verb Meaning in Understanding AS-
alternations ___________________________________________________________ 4 2.4 Argument Classification: Thematic / Semantic Roles _______________________ 5 2.5 External/Internal Argument Distinction, Unaccusativity and the Unaccusative
Hypothesis ___________________________________________________________ 6 2.6 Impersonal Constructions: Expletives, Passives and SwarmAlt _______________ 7 2.7 Aspectual Distinctions: Telic / Atelic ___________________________________ 13 2.7.1 Testing for Situation Types: Telicity ________________________________ 13 3. The Swarm Alternation in English ________________________________________ 15 3.1 The Swarm Alternation in English _____________________________________ 15 3.1.1 Optional / Obligatory Arguments __________________________________ 16 3.2 SwarmAlt Verbs, Telicity and Holistic Effect _____________________________ 16 3.2.1 The Relation of the Swarm Alternation to Spray/Load __________________ 16 3.2.2 SwarmAlt Verbs are Atelic ________________________________________ 17 3.2.3 The ‘Holistic Effect’ in L-subject sentences ___________________________ 18 3.3 Some Observations: Semantic Properties of the L-subject variant ____________ 19 3.3.1 Dowty’s (2000) L-subject Observations: “Five Characteristics” ___________ 19 3.3.2 Dowty’s (2000) Five Meaning Classes_______________________________ 20 3.3.3 Restrictions on the With-phrase in the Marked Variant _________________ 21 3.3.4 The Literal / Metaphorical meaning distinction and A-subject vs L-subject __ 22 3.4 Interim Summary __________________________________________________ 24 4. The Swarm Alternation in Norwegian, and Comparison with English _____________ 25 4.1 The Det-subject Pattern _____________________________________________ 25 4.2 The Norwegian Swarm Alternation: three patterns rather than just two? ______ 27 4.3 Telicity and ‘Holistic Effect’ in Norwegian SwarmAlt _______________________ 28 4.3.1 The ‘Holistic Effect’ _____________________________________________ 29
4.3.2 Telicity _______________________________________________________ 29 4.4 Optionality _______________________________________________________ 30 4.5 Norwegian SwarmAlt, Verbs and Verb Classes ___________________________ 33 4.6 The Literal / Metaphorical meaning distinction in Norwegian SwarmAlt ________ 36 4.7 Restrictions on the av/med (by/with) PP in the Marked Variant ______________ 37 4.8 SwarmAlt with Verb Phrases denoting Directed Motion_____________________ 37
4.8.1 Directed Motion Verbs in Norwegian SwarmAlt: does Norwegian SwarmAlt involve Locational as well as Directional PPs? _____________________________ 38 4.8.2 An English Version of the Norwegian Det-variant? _____________________ 40 4.9 SwarmAlt as an Unergative/Unaccusative Alternation? _____________________ 42 4.10 Interim Summary _________________________________________________ 45 5. Concluding Remarks ___________________________________________________ 47 5.1 Main Points and Tentative Conclusions _________________________________ 47 5.2 Issues for Future Research___________________________________________ 47 List of References _______________________________________________________ 49 Abstract_______________________________________________________________ 51 Appendix A ____________________________________________________________ 52 Appendix B ____________________________________________________________ 53
List of Tables
Table 1 Outline of Thematic Roles ___________________________________________ 6
1. Introduction
1.1 Topic and Approach
The main focus of this thesis is to do a comparative analysis of a specific argument alternation - the swarm alternation - in English and Norwegian. This is first of all
interesting in its own right, as to my knowledge no previous work has been done on this particular alternation in Norwegian. The English swarm construction, however, has been studied more thoroughly (E.g. Salkoff (1983); Levin (1993); Dowty (2000); Rowlands (2002)). In light of key properties discovered in the English SwarmAlt, I intend to primarily investigate what appear to be Norwegian equivalents, in order to shed light on two things: (i) the nature of the phenomenon in Norwegian and (ii) English-Norwegian similarities and differences in this domain. This may furthermore lead to greater
understanding of the phenomenon in Norwegian, but also a greater understanding of the relationship between English and Norwegian, and English-Norwegian similarities and differences in this domain.
For the main part my study will be conducted within the generative grammar framework, and my main focus is to answer descriptive questions such as ‘are the (non)alternating verbs the same in the two languages’ and ‘what are the Norwegian SwarmAlt
(syntactic/structural) patterns’. Since Dowty (2000) has discovered several aspects regarding the swarm alternation in English (especially regarding the L-subject form), there are also several focus areas which are related to his observations. These include examining the verb classes associated with SwarmAlt; the English with-phrase
restrictions in the corresponding Norwegian av/med-PP, and the distinct syntactical patterns associated with literal vs. metaphoric interpretations. Most space will however be devoted for the structural pattern in the marked variant as well as the question of whether Norwegian SwarmAlt involves locational as well as directional PPs, as it appears that it is in this regard that the differences are most notable. Last but not least, I am also posing the question of whether the swarm alternation is an unergative/unaccusative alternation, contrary to previous research which has analysed it as either unaccusative (e.g. Levin (1993)) or unergative (e.g. Rowlands (2002)).
As I am mostly focusing on establishing basic aspects of the hitherto not investigated alternation in Norwegian based on what has already been done in English, the more complex explanations and/or questions that arise may not be easily answered within the scope of a master’s thesis. However, I am aiming for providing a thorough description, which might lay a foundation for more comprehensive subsequent study and research.
1.2 Main Points
Through analysing the data examined in this thesis, it has been established that the swarm alternation in English and Norwegian is in many respects highly similar. Thus, much of the previous research on the swarm alternation in English has been
substantiated when exploring similar aspects in Norwegian. The similarities are to a great extent related to the structure of the (unmarked) A-subject pattern, (non)alternating verbs, and the semantic verb classes appearing in the alternation, but the data also illustrate some apparent differences between the alternation in the two languages. The first notable difference, which also is one of the main points in the thesis, is that the Norwegian swarm alternation involves two marked patterns, sometimes giving rise to a three-way alternation: an impersonal (det-subject) pattern, which is specific to
Norwegian, and L-subject, which is equivalent to the marked pattern in English
SwarmAlt. Norwegian SwarmAlt thus involves a slightly more complex alternation pattern than what has been noted for English. This is also connected to one of the biggest
differences between English and Norwegian; the use of an impersonal sentence pattern for one of the special variants in Norwegian.
Furthermore, I will argue that the swarm alternation contains a signature property, which is found in the alternation in both languages; ‘agent’ realised in a with- / av/med-PP with special restrictions (this observation is due to C. Wilder (p.c.)). This is moreover of special importance regarding a second major difference, which relates to how Norwegian SwarmAlt involves a type with directional PPs (directed motion verbs), and although English SwarmAlt also has motion verb type, it is much more restricted (Wilder, 2019).
1.3 Structure
Including the introduction chapter 1, the thesis is divided into five main sections. The next section, 2 (Background), is where I introduce the theoretical frame for the thesis, the basic patterns of the swarm alternation in both English and Norwegian, as well as explain relevant terms and phenomena. Chapter 3 (The swarm alternation in English) focuses on presenting and explaining some of the main results of previous research regarding the swarm alternation in English, which will include elaborating on the SwarmAlt patterns in English, as well as discussing specific characteristics of the
alternation such as the holistic effect in the L-subject variant, that the swarm alternation is atelic, as well as the semantic properties of the L-subject variant observed by Dowty (2000). In chapter 4 (The swarm alternation in Norwegian and comparison with English) I will use the main results of previous research on the swarm alternation in English as a starting point for exploring and discussing equivalent aspects of the swarm alternation in Norwegian. I will start by establishing the basic pattern of the alternation, and although the structure is generally largely similar to chapter 3, there are a few differences as a result of important observations regarding the swarm alternation in Norwegian. The first part of the chapter focuses mostly on comparing, explaining and discussing aspects that are already observed in English SwarmAlt, followed by presentation and discussion of new data. In chapter 5 (Concluding remarks), some tentative conclusions will be drawn and there will be given some suggestions for future research.
2. Background
In the sections below I will outline and explain central aspects related to the theoretical framework of this thesis, in addition to a first introduction to the basic form of the swarm alternation in English and Norwegian. I will moreover explain terminology that are
relevant and/or useful when discussing argument alternations in general, though especially regarding the main aims of this thesis and what will be discussed in the subsequent chapters.
2.1 Argument Alternations
The overarching topic and background for the master’s thesis is a phenomenon referred to as argument structure alternations, which according to Dowty (2000:111) has been a highly debated topic among linguists since the very beginning of generative grammar.
Levin and Rappaport state that the term refers to how some verbs show “alternate
expressions of their arguments” (Levin & Rappaport, 2005:5) and furthermore represents one of several instances of “the larger phenomenon of multiple argument realization”
(Levin & Rappaport, 2005:188-9). This refers to the way most verbs allow their
arguments to be expressed in more than one way (Levin & Rappaport, 2005:5), and as in the swarm alternation illustrated in (1), the arguments of the verb can be realized either in a NP1 V PPloc (in/on) NP2 structure or a NP2 V PPwith NP1 structure.
(1) a. Bees swarmed in the kitchen. b. The kitchen swarmed with bees.
(2) a. He gave the flowers to her. b. He gave her the flowers.
(3) a. She sprayed water on the wall. b. She sprayed the wall with water.
(4) a. She hit the door. b. She hit at the door.
In addition to SwarmAlt (1), which will be the main focus of this thesis, several different types of alternations have been realized in the literature, e.g. the dative alternation (2), locative / spray/load alternation (3) and the conative alternation (4). Levin & Rappaport state that verbs that are similar in meaning do not always show the same alternations, yet despite this apparent idiosyncrasy argument alternations can be extended to new verbs (Levin & Rappaport, 2005:5). They furthermore claim that “a verb’s root has a major part to play in determining which forms of multiple argument realization it might show” (Levin & Rappaport, 2005:5).
What should additionally be noted is that in many argument structure alternation
patterns, one of the two alternants can be identified as being more “special” or “marked”.
As stated by Rowlands (2002:6), one of the variants often involves an additional sense, and this variant is often more restricted, not as common as or even more complex than the other variant, which is more basic. Certain properties can distinguish which of the two alternants is more special, and this can be connected to i) morphological marking on the verb, ii) argument linking and iii) use of special (grammatical) preposition to mark an argument. These properties may differ with respect to which alternation is in question, however, with regard to the swarm alternation in English one can especially detect which of the variants is more special or marked due to ii) and iii), e.g. that the AGENT appears as the subject in the unmarked variant whereas LOCATION is the subject in the other, which is a more special argument realization. As will be presented in the next section, the unmarked variant furthermore involves a (locative) in/on-PP, whereas the marked variant has a with-PP, where with seems to be purely grammatical as opposed to for instance with-instrument.
2.2 Introduction of Basic Structure: The Syntactic Pattern(s) in Question
The alternation in question, especially in English, has been investigated by a number of linguistic scholars (e.g. Salkoff (1983); Levin (1993); Dowty (2000); Rowlands (2002);
Hoeksema (2009)). The Norwegian counterpart, on the other hand, has to my knowledge not been the subject of previous investigation. However, based on the discussion in the subsequent chapters, the syntactic patterns of the swarm alternation and its variants in English and Norwegian will be assumed to be as presented below.
English pattern:
(5) a. NP1 V PPLOC (in/on) NP2 (A-subject variant) b. NP2 V PPwith NP1 (L-subject variant) Norwegian pattern:
(6) a. NP1 V PPLOC (i/på) NP2 (A-subject variant)
b. NP2 V PPav/med NP1 (L-subject variant)
c. Det V PPav/med NP1 PPLOC NP2 (Det-subject variant) The swarm alternation is an alternation found with (atelic) intransitive verbs (Dowty, 2000:114), and as the patterns outlined above illustrate, the NP1 V PP1LOC NP2 structure in both English and Norwegian involves a locative PP which is headed by the prepositions in/i or on/på. However, whereas the NP2 V PP2 NP1 structure consistently involves a with-PP in English, Norwegian sometimes allow the use of av (of/by) in addition to med (with). As will be discussed and argued in chapter 4, Norwegian SwarmAlt involves two marked patterns; a special det-subject (impersonal) pattern in addition to the (somewhat more restricted) NP2 V PP2av/med NP1 structure. Example sentences of the swarm
alternation are given for both English and Norwegian in (7) below.
(7) a. Stars glittered on the sky.
b. The sky glittered with stars.
c. Stjerner glitret på himmelen.
Stars glittered on sky.the ‘Stars glittered in the sky’
d. Det glitret av stjerner på himmelen.
It glittered by stars on sky.the ‘The sky glittered with stars’
e. Himmelen glitret av stjerner.
Sky.the glittered by stars ‘The sky glittered with stars’
2.3 The Lexicalist Approach and the Role of Verb Meaning in Understanding AS-alternations
Over the years of studies regarding argument alternations, several scholars have
pursued the attempt to pin down what factors govern which verbs alternate and which do not. Several theories of argument realization have been developed and a central tenet among a number of linguistic scholars (e.g. Levin (1993); Levin & Rappaport (1995, 2005)) is that the verb and verb meaning is significant to argument linking (Levin &
Rappaport, 2005:2-3). As further stated by Levin and Rappaport, it has been known for a
long time that verbs group together in classes with similar meanings that are
semantically identifiable, and the fact that these classes show characteristic patterns of argument realization suggests that there might be a connection between these patterns and the semantic properties of each class (Levin & Rappaport, 2005:1-2).
Such a view of argument realization and argument alternations is connected to the lexicalist approach, which is a theoretical perspective/standpoint within modern
generative linguistics. Within the lexicalist approach, argument structure and realization of a verb’s argument is assumed to be driven by “information registered in a structured lexical entry for that verb” (Levin & Rappaport, 2005:186). Other scholars take a different stand however, arguing that most of the determinants of argument realization are not strictly lexical, rather, they suggest that certain syntactic configurations are themselves the bearers and determiners of certain meaning components (Levin and Rappaport, 2005:6). According to Levin (1993:5), however, a more in-depth
investigation of the nature of lexical knowledge confirms that different aspects regarding syntactic behaviour are tied to verb meaning; verbs behaving in similar ways are
expected to reflect and share meaning components.
For instance, investigations of the verbs break, cut, hit and touch (e.g. Filmore (1967), Guerssel, Hale, Laughren, B. Levin and White Eagle (1985), Hale and Keyser (1986, 1987), and Laughren (1988)) show that although they are all transitive two-argument verbs, they do not have much else in common (Levin, 1993:5-6). They furthermore differ with regard to whether or not they participate in a number of argument structure
alternations, such as the middle construction, the conative construction and the body- part possessor ascension alternation. According to Levin (1993:7-10), their behavioural pattern is connected to the specific meaning components “contact”, “change” and
“motion”, which are the aspects of meaning that “serve minimally to distinguish the verbs participating in the different alternations” (Levin, 1993:10).
That verb meaning is essential to syntactic behaviour is however a somewhat
controversial hypothesis which has several challenges, such as for instance answering the questions of to what extent does the meaning of verbs determine their syntactic behaviour and how does one determine which meaning components that are relevant (Levin, 1993:12-13). There are furthermore several researchers who have argued for the rejection of this hypothesis (Levin, 1993:12). Nevertheless, as Levin further states, “ […]
the examples above and examples cited in other works illustrate, suggests that the ties between a verb’s meaning and syntactic behaviour cannot simply be ignored” (Levin, 1993:12).
2.4 Argument Classification: Thematic / Semantic Roles
When it comes to argument realization, the argument structure of verbs and the different arguments involved in a verb’s syntactic representation, these can be assigned specific roles. This is a widely adopted form of lexical semantic representation (Levin &
Rappaport, 2005:35) and as stated by Saeed (2016:149), such roles are referred to with a number of different labels in semantics, including thematic roles (e.g. Dowty 1986, 1989, 1991), thematic relations (Jackendoff 1972), deep semantic cases (Fillmore 1968), participant roles (Allan 1986) and semantic roles (Givón 1990). For instance, the verbs break and put might be associated with the semantic roles “Agent, Patient” and “Agent, Theme, Location” (Levin & Rappaport, 2005:35), as illustrated in the examples below.
The table below briefly outlines central thematic roles largely based on Saeed
(2016:150). Thematic roles should however not be mixed up with subject, object etc.
(i.e. grammatical roles of arguments) (Haegeman & Guéron, 1999:25-26).
(8) a. She broke the chair.
AG PAT
b. I put the book in the bedroom.
AG TH LOC Table 1 Outline of Thematic Roles
AGENT: Initiator of action, often with intention.
PATIENT: Entity affected by some action, often change of state.
THEME: Entity moved by the action or whose location is described.
EXPERIENCER Entity has awareness of the action/state but is not in control of it.
BENEFICIARY: Entity benefitting from the action.
INSTRUMENT: The means by which an action is performed.
LOCATION: The place wherein something takes place or is situated.
GOAL: The entity toward which something moves, literally or metaphorically.
SOURCE: The entity from which something moves, literally or metaphorically.
Despite being involved in a wide range of linguistic contexts, the thematic role approach is subject to much disagreement, and involves several challenges and/or issues raised by a number of scholars. As stated by Dowty (1991:548-9) even the most familiar roles are not agreed upon, and new role labels have also emerged along the way. Some issues have to do with how finely each role should be divided and/or defined, as well as the potential for dual role assignment (Dowty (1991:553-556); Levin & Rappaport (2005:38- 42)). Last but not least, it has also been noted that the thematic role approach does not give a sufficient account of argument alternations, e.g. with regard to semantic
differences between the two variants (Levin & Rappaport (2005:154); Arad (2006)), such as in spray/load and swarm. As stated by Rambøl (2010:26) however, thematic roles are nevertheless useful terms for discussing argument alternations.
2.5 External/Internal Argument Distinction, Unaccusativity and the Unaccusative Hypothesis
The Unaccusative Hypothesis was first posed by Perlmutter (1978), and later adopted by Burzio (1986), and the proposal is that there are two subclasses of intransitive verbs, unergative verbs and unaccusative verbs, each associated with a distinct (underlying) syntactic pattern (Levin & Rappaport, 1995:2). The unaccusative intransitives act as if they have an underlying object/theme which surfaces as the subject in the sentence (i.e.
have an internal but no external argument) (Butt, 2006:39). In P&P, regarding
“structural Case”, it is assumed that the argument NP moves to the structural subject position in order to receive nominative Case (Butt, 2006:60). The single argument of unergative verbs on the other hand, is an underlying subject (i.e. have an external argument but no internal argument) (Levin, 1993:13).
Moreover, whether the verbs participating in the swarm alternation are unergative or unaccusative is a question posed by several scholars (e.g. Levin (1993); Rowlands
(2002)). Levin (1993:50) treats swarm as involving unaccusative verbs, which is
connected to her proposal of the (transitive) spray/load alternation and the (intransitive) swarm alternation as both being subtypes of the locative alternation. In her analysis, Levin (1993:50) states that the forms of the alternation manifested by spray/load and swarm can be viewed as a single alternation if the intransitive swarm verbs are analysed as unaccusative.
Rowlands (2002:22-23) on the other hand, largely basing her claims on the
‘causativisation test’, argues that swarm verbs are to be analysed as unergative. It has been claimed that unergatives resist causativisation, and based on sentences like in (9) below illustrating that swarm cannot be causativised, Rowlands (2002:23) concludes that swarm verbs are unergative.
(9) a. ?John swarmed the garden with bees.
b. *John swarmed the bees in the garden.
c. *Paula buzzed the jar with flies.
d. *Paula buzzed the flies in the jar.
The formation of pseudopassives (prepositional passives) can also be applied as an unaccusativity diagnostic, as it has been argued by e.g. Perlmutter and Postal (1984) and Levin and Rappaport (1995) that only unergative verbs, based on sentences as in (10) below, are possible in this construction in English (cited in Kuno & Takami, 2004:137).
Although the results from various diagnostics for unaccusativity generally should be considered with care since there appears to be counterexamples to most diagnostics, including the pseudopassive construction (cf. Kuno & Takami, 2004:145-147), acceptable formation of pseudopassives with e.g. verbs related to the swarm alternation, might point in the direction that they are unergative.
(10) a. *Boston was arrived in late at night.
b. The bed was slept in by Napoleon.
Furthermore, which will be further elaborated in the following section discussing impersonal constructions, having an expletive in the subject position has also been proposed as a way to distinguish unaccusatives from unergatives in both English and Norwegian. The there-construction came to be seen as an ‘unaccusativity diagnostic’ in English in the 1980-90s, and as stated by Sveen, Åfarli (1992) takes the possibility of having an expletive in subject position to be a defining characteristic of unaccusativity (cited in Sveen, 1996:134).
2.6 Impersonal Constructions: Expletives, Passives and SwarmAlt
Both English and Norwegian have (active) sentences which contains an expletive, a non- referential element, in the subject position – it/there in English and det in Norwegian – which appears to merely function as a placeholder for the ‘real’ subject and does not contribute anything to the semantics of the sentence (Haegeman & Guerón, 1999:42). In Norwegian there additionally exists options to form passives with impersonal det.
However, the range of possibilities for both expletive and passive sentences is much wider in Norwegian compared to English (Sveen (1996:97); Åfarli (1992:78)), and the resemblance to one of the marked variants of SwarmAlt, namely the det-variant, and these sentences, is striking. The different types of impersonal constructions in
Norwegian, several of which are identified by Sveen (1996), are discussed and exemplified below.
Type 1: Clausal extraposition sentences: [SU-clause] V … / Det V … [SU-clause]
(11) a. At Rosenborg tapte overrasket alle.
That Rosenborg lost surprised everyone ‘That Rosenborg lost surprised everyone’
b. Det overrasket alle at Rosenborg tapte.
It surprised everyone that Rosenborg lost ‘It surprised everyone that Rosenborg lost’
c. That Manchester United lost surprised everyone.
d. It surprised everyone that Manchester United lost.
Clausal extraposition sentences with an expletive/placeholder subject is found in
Norwegian as well as English. Although SwarmAlt does not involve verbs that take clausal complements, clausal extraposition sentences should also be mentioned, since they represent yet another case where an argument of the main verb/predicate alternates between realisation as preverbal subject in the neutral pattern (11a+c) and realisation in postverbal position with expletive det (it) in the preverbal subject position in the special pattern (11b+d) (Wilder, 2018a:43).
Type 2: det + V + (postverbal) subject (12)
a. Noen barn lekte på plenen. b. Det lekte noen barn på plenen.
some kids played on lawn.the there played some kids on lawn.the ‘Some kids were playing on the lawn’ ‘Some kids were playing on the lawn’
c. Noen kommer / Noen banker på døra. d. Det kommer noen / Det banker noen someone is coming / someone knocks on på døra.
door.the there comes someone / there knocks ‘Someone is coming / Someone is on door.the
knocking on the door’ ‘Someone is coming / Someone is knocking on the door’
The first type of impersonal actives discussed by Sveen (1996) is one where the subject NP either is in the subject position (12a) or the subject NP is postponed and the expletive det is in subject position (12b). In this structure the agent argument, which is usually accounted for as an “external” argument, is realized postverbally in the (internal) direct object position (Sveen, 1996:157-158). This impersonal construction also exists in English, and the English expletive for type 2 is there. However, type 2 impersonal
constructions is notably much less frequent in English than in Norwegian. Apart from with the verb be (existential sentences), very few verbs (verbs of existence and appearance) are thought to be possible in this construction in English (Wilder, 2018a:41-42).
Related to this is furthermore how the there-construction came to be seen as an
‘unaccusativity diagnostic’ in English in the 1980-90s. Although it was first claimed that only be is possible in this construction, several scholars (e.g. Burzio (1986); Levin &
Rappaport (1995)) observed that there was a distinction between the copular verb be as well as certain (main) verbs of existence and appearance (e.g. exist, appear) on the one
hand, and other main verbs like play, dance and sing on the other (Kuno & Takami, 2004:31-2). Following this observation and based on sentences like in (13)-(14) below, it was claimed that only (unaccusative) verbs of existence and appearance are possible in the there-construction (e.g. Burzio (1986); Levin & Rappaport (1995)), whereas no unergatives (or transitives) are (Kuno & Takami, 2004:32). The there-sentences containing unaccusative verbs, e.g. a policeman appeared, were analaysed as having a surface subject which in D-structure was an underlying direct object. Whereas the underlying object normally raises to SpecIP to become the surface subject (13a), the underlying object in there-sentences stays in its D-structure position, and an expletive is inserted in SpecIP (13b).
(13) a. The policeman appeared.
b. There appeared a policeman.
c. There arose a storm
(14) a. ??There knocked someone (at the door) b. ??There played some children (in the field)
Kuno and Takami (2004) however, provide several counterexamples to the
unaccusatives-only claim for the English there-construction, illustrating that the there- construction as an unaccusative diagnostic is flawed. Sentences like those in (15) below are presented by Kuno and Takami (2004:40-41) in order to illustrate and discuss their points about the appearance of unergatives (15a) and transitives (15b) in the there- construction, despite previously claimed otherwise.
(15) a. There once ruled a king who had no ears.
b. Then, all of a sudden, there reached her ear the sound of angel voices.
The same basic claims as for the there-construction were furthermore made concerning Norwegian det-sentences of type 2 (e.g. Åfarli, 1992:89-90), suggesting that only unaccusatives occur in this construction. However, it was discovered that there are several Type 2 examples with verbs that are not unaccusative but which comes out as unergative by other diagnostics, that can occur in this construction, e.g. leke (play) as in (12b) above, banke (knock) (12d) and arbeide (work). Though Sveen (1996:133-4), who challenges this analysis, states that it appears that all the writers dealing with
(Scandinavian) impersonal actives (e.g. Åfarli (1992)) share the same underlying premise:
[…] that (seemingly) unergative verbs occurring in impersonal actives cannot be what they seem, they must have somehow taken on unaccusative characteristics and have an internal argument, they cannot be straightforward unergatives with an external argument.
In the framework of the classic syntactic analysis of unaccusativity there is no
straightforward account for these phenomena, as is also pointed out by Sveen (1996), and the fact that verbs of both (unacc+unerg) types can occur in the construction thus suggests that just as the there-construction cannot be said to necessarily diagnose unaccusativity in English, neither can Åfarli’s “expletive test” in Norwegian (Sveen, 1996:136). According to Sveen, Åfarli’s (1992) claims partly stem from neglecting the crucial fact that in Norwegian “intransitive verbs in general can occur in impersonal active
sentences, regardless of whether the verb is unaccusative or unergative” (Sveen, 1996:140).
Type 3: weather-verb pattern: det + V (16) a. Det blåser/regner/snør.
It is blowing/raning/snowing
‘It is windy/raining/snowing’
b. *Himmelen snør.
Sky.the is snowing c. Det hosta hele tiden.
It coughed all time.the
‘Someone was coughing all the time’
d. Pasientene hosta hele tiden.
Patients.the coughed all time.the
‘The patients were coughing all the time’
As stated by Sveen (1996:215), the verbs associated with the weather-verb pattern belong to what he refers to as “the perceptive class”. However, unlike the perceptive class more generally, the type 3 weather-verb pattern most often seem to have no argument at all (16a-b) (Sveen, 1996:215). This also serves to distinguish weather- verbs from “weather expressions” more generally. Compared to type 2 then, type 3 involves a verb which lacks an argument. The Det+V pattern does not alternate with another pattern, unlike Type 2, where the structure NP V alternates with Det V NP.
English also has a similar impersonal construction with e.g. weather expressions.
However, it is crucial to note that verbs like e.g. pour as in (17) describe weather
phenomena, and unlike the weather-verbs of Norwegian type 3 impersonal actives these verbs do take an argument NP. They show a special alternation pattern wherein the neutral pattern involves the structure NP V (PPdir) (17c) and the marked pattern involves the structure it V (PPdir) PPwith (17b). It is furthermore interesting to note that the marked pattern only seems possible with weather expressions in English, just as type 3 sentences seem limited to weather-verbs in Norwegian.
(17) a. It is raining.
b. It is pouring down with rain (in Trondheim).
c. Rain is pouring down (in Trondheim).
Type 4: ‘perceptive construction’: SU V / Det V
(18) a. NP V neutral
b. det V marked
In what Sveen (1996:210) refers to as the “perceptive construction” the verbs either appear without an argument, as in (18), or the argument appears inside a PP (“oblique realization”). In the neutral pattern in (18a) the verb appears in the structure NP+V, whereas in the marked pattern the structure is det + V, wherein it seems as if the verb
“loses” its subject argument. Comparing the sentences with the verbs banke (knock) and brenne (burn) in (19a-b) and (19c-d) below, the impersonal sentence pattern has the expletive in the subject position rather than the NP someone or house.
(19) a. Noen banker på døra / Noen går i trappa neutral Someone is knocking on door.the / someone is walking in stairs.the
‘Someone is knocking on the door / someone is walking in the stairs’
b. Det banker på døra / Det går i trappa marked It is knocking on door.the / it is walking in stairs.the
‘Someone is knocking on the door / is walking on the stairs’
c. Huset brenner.
House.the is burning ‘The house is on fire’
d. Det brenner.
It is burning
‘Something is on fire’
According to Sveen (1996:216), the verbs appearing in the perceptive construction share a common semantic characteristic, a “property of perceptibility”, meaning that they refer to some event which is directly perceptible to our senses, either heard, seen, felt or smelt, literally as well as metaphorically (Sveen, 1996:211). The perceptibility property is moreover claimed to be connected to the overall acceptability of impersonals (Sveen, 1996:216). Sveen claims that whether the occurrence of a verb in an impersonal active is acceptable or not has to do with whether the activity denoted by the verb is readily perceptible (Sveen, 1996:216). This echoes the ideas of Dowty (2000) with regard to (verbs appearing in) the English SwarmAlt, and indeed among the different cases of the
“oblique realization” instances of the “perceptible construction” presented by Sveen (1996), some are definitely (Norwegian) SwarmAlt cases (Wilder, 2020). Some of the examples presented by Sveen (1996:213-214) are given in (20) below and in chapter 4 I will make the proposal that this represents a new, separate det-sentence type (type 5).
(20) a. Det krydde av/med unger i nabolaget.
it swarmed by/with kids in neighbourhood.the
‘The neighbourhood was swarming with kids’
b. Det surret av stemmer i rommet.
it buzzed by voices in room.the
‘The room was buzzing with voices’
According to Sveen, the instances of the perceptibility construction with “oblique
realization” are associated with expression of the abundance of something, which in turn makes it very much heard/seen/felt, i.e. perceived (Sveen, 1996:213). The PP in these cases is mostly headed by av (of/by), but sometimes the PP can also be headed by med (with) (Sveen, 1996:213). Although these are definitely SwarmAlt examples, Sveen does not mention or refer to any literature on (English) SwarmAlt. Sveen’s work precedes Dowty (2000), however despite citing e.g. Levin (1993) in his work, there is no reference to any discussion of English SwarmAlt. Similarily, Dowty and others writing after 1996 also appear unaware of Sveen’s work.
There are however more or less subtle aspects which serve to distinguish type 3 and 4 from the SwarmAlt cases (Wilder, 2020), such as the (‘agent’) av/med-PP. As will be illustrated and discussed in the section (4.1 ‘The det-subject pattern’) and (4.4
‘Optionality’), what I will argue to be “the SwarmAlt pattern” most often involves the
expression of the ‘Agent’ in an av/med-PP, whereas as illustrated in (21) below, type 3 and 4 do not.
(21) a. Det banker (*av/med noen) på døra.
It knocks (by/with someone) on door.the ‘Someone is knocking on the door’
b. Det går (*av/med noen) i trappa.
It walks (by/with someone) in stairs.the ‘Someone is walking on the stairs’
c. Det regner (*av/med regn).
It rains (by/with rain) ‘It is raining’
In addition to the various types of impersonal actives presented above, the placeholder pronoun det is also quite frequently used to form passives in Norwegian (Wilder,
2018a:48), though it is especially the possibility to form impersonal passives which is of importance. As stated by Sveen (1996:52), if a verb can be passivized in Norwegian, an impersonal passive can also be formed. With transitive verbs the passive verb can occur in two patterns, one where the subject of the sentence is a derived subject (similar to English) and one impersonal pattern with an expletive subject, i.e. the expletive subject is an option (Sveen, 1996:53). However, in passives formed from intransitives, the impersonal patterns is the only pattern, i.e. the expletive subject is a “syntactic necessity” (Sveen, 1996:53). The result is that in Norwegian the subject position of a passive contains either an expletive or an NP (as a result of NP-movement), in order to meet the demand for a subject by the Extended Projection Principle, whereas English almost altogether lacks impersonal passives (and passives from intransitive verbs) (Åfarli, 1992:80).
The sentences in (22b-d) have no direct counterpart in English, though by comparing the impersonal passives with SwarmAlt, one can also see that SwarmAlt looks a lot like passivisation in Norwegian (This observation is due to C. Wilder (p.c.)). Whereas the agent in passives in English can be expressed in a by-phrase, av is the special
preposition which marks the agent in Norwegian (Wilder, 2018a:53-54). This preposition is also used to mark the ‘Agent’-phrase in Norwegian SwarmAlt. Although the av-phrase is sometimes possible to express in impersonal passive constructions (22b+d), it usually sounds odd, especially with impersonal passives of intransitive verbs (Sveen, 1996:55).
(22) a. Slike feil blir oppdaget av våre arbeidere hver dag.
Such mistakes becomes discovered by our workers every day
‘Errors of this kind are discovered by our workers every day’
b. Det blir oppdaget slike feil av våre arbeidere hver dag.
It becomes discovered such mistakes by our workers every day
‘Errors of this kind are discovered by our workers every day’
c. Det ble arbeidet i går.
It became worked yesterday
‘People were working yesterday’
d. Det blir jobbet kontinuerlig (av våre ansatte).
It becomes worked continuously (by our employees)
‘Our employers were continuously working’
2.7 Aspectual Distinctions: Telic / Atelic
As stated by Saeed (2016:118), verbs can be characterized in terms of a number of different situation types, such as static/dynamic; durative/punctual; telic/atelic. Whether a (dynamic+durative) verb is telic or atelic is often discussed as part of the
characteristics of argument structure and different argument alternations (e.g. Levin &
Rappaport, 2005:103), such as the swarm alternation which is said to involve verbs that are “almost all atelic” (Dowty, 2000:114). The telic/atelic distinction is furthermore an important semantic distinction regarding verbs in different dynamic situation types (Saeed, 2016:116-118). Whereas telic refers to processes where the event is seen as having a natural endpoint and an “implied boundary for when the process will be over”
(Saeed, 2016:117), such as build in (23d) below, atelic refers to processes which can continue indefinitely, i.e. there is no implied boundary for when the process will be over (Saeed, 2016:117). If the processes in (23a) and (23c) are interrupted at any point, it would still be correct to say (23b), but it is not necessarily correct to say (23d) (Saeed, 2016:117).
(23) a. Maddy was gazing at the stars. b. Maddy gazed at the stars. (atelic) c. Martin was building a house. d. Martin built a house. (telic) Although differing between (a)telic events might seem fairly clear-cut in theory, Saeed (2016:117) further states that combining inherently telic or atelic verbs with other sentence elements (e.g. adverbials) can alter the situation type and result in a different aspect (Saeed, 2016:117-119). Thus, since for instance an atelic verb can be interpreted as a telic situation type, depending on other elements in the sentence (24b), determining whether a verb is telic or atelic is not always an easy task.
(24) a. I was running. (atelic)
b. I was running in the Trondheim Marathon. (telic) 2.7.1 Testing for Situation Types: Telicity
Although the task of determining whether an event is telic/atelic is not always as clear cut, there are certain tests that can be helpful to decide what type a clause belongs to. A common test suggested by Dowty (1979) in order to distinguish between telic and atelic situation types is by means of various temporal adverbial expressions (Dowty (1979) in Saeed (2016:121)). In (a period) is a temporal adverbial which only occurs with telic situation types, whereas the durational adverbial for (a period) occurs with atelic situation types. This is illustrated in the sentences below.
(25) a. Martin built a boat (in half an hour). (telic) b. Maddy gazed at the stars (*in half an hour). (atelic) (26) a. Martin built a boat (for half an hour). (telic)
b. Maddy gazed at the stars (for half an hour). (atelic)
3. The Swarm Alternation in English
In this chapter the main focus is to present and explain some of the main results of previous research regarding the swarm alternation in English (e.g. Levin (1993); Dowty (2000); Rowlands (2002); Hoeksema (2009)). This will furthermore be part of setting the context for exploration of Norwegian SwarmAlt in the next chapter (ch. 4).
3.1 The Swarm Alternation in English
In this section the English SwarmAlt pattern will be presented, which includes the syntactic patterns of the two variants associated with the alternation as well as the semantic arguments of the verb. Furthermore, among the central aspects in this section is also to identify and explain why the with-variant is the “special” pattern, as well as describe the optional / obligatory status of the PPs involved in the alternation patterns.
Based on sentences such as those in (27)-(28), the structure of English SwarmAlt can be illustrated as in (29) below.
(27) a. NP1 V PPLOC NP2
Insects are crawling on the terrace. (A-subject) b. NP2 V PPwith NP1
The terrace is crawling with insects. (L-subject) (28) a. Birds are swarming in the garden.
b. The garden is swarming with birds.
Syntactic pattern:
(29) a. NP1 V PPLOC (in/on…) NP2 (A-subject variant)
b. NP2 V PPwith NP1 (L-subject variant)
Dowty (2000) refers to the unmarked variant as the A(gent)-subject form and the
marked variant as the L(ocation)-subject form, though, in many cases the term ‘Agent’ is not appropriate. For instance, with many verbs, the “agent” argument referent does not
“act with volition” in the situation denoted, and with a number of verbs, the referent can or must be inanimate. I will nevertheless follow the same notation as Dowty (2000), for the sake of having a label. The two different ways of expressing the arguments of the verb in the English swarm alternation are illustrated in (27)-(28) above.
The swarm alternation is moreover an alternation found with intransitive verbs (Levin, 1993:54), which are “almost all atelic” (Dowty, 2000:114). It furthermore constitutes alternation of arguments in subject and complement positions, and verbs such as swarm may be realized in the syntactic pattern NP1 V PPLOC NP2 or NP2 V PPwith NP1. The
structure in (29a) illustrates the unmarked variant, the A-subject form, where the AGENT is the subject and the LOCATION argument appears in a PPLOC, whereas the structure in (29b) illustrates the marked variant, the L-subject form, where the LOCATION is the subject and the AGENT appears in a with-PP (Dowty, 2000:112).
Of the two SwarmAlt patterns, the L-subject/with-variant is argued to be the more
“special” pattern (Dowty (2000:120); Rowlands (2002:31)). Reasons why the L-subject form containing the with-PP is special is related to the fact that although verbs that appear in the L-subject variant generally have a matching A-subject variant, there are many (atelic activity) verbs which appear in the A-subject pattern though have no L-
subject counterpart (Rowlands, 2002:31), e.g. Bees flew in the garden / *The garden flew with bees. The fact that there are several verbs which are acceptable in the A-
subject variant but not the with-variant illustrates a “marked difference” between the two variants according to Rowlands (2002:31). This asymmetry between the two variants is part of the reason for arguing that the L-subject/with-variant is the more special frame, and the asymmetry is especially underpinning Rowlands (2002:31) argument that more attention should be devoted to the with-variant.
A related pattern is moreover found with adjectives (as opposed to verbs). As part of his fairly detailed investigation of the swarm alternation, Salkoff (1983) introduces two adjectival frames involving adjectives derived from verbs as well as underived adjectives in addition to the two SwarmAlt frames (Salkoff (1983) cited in Rowlands (2002:16).
These are illustrated in (30) below, though, most of these instances have no acceptable A-subject variant (Salkoff (1983) cited in Rowlands (2002:17). These related adjectival patterns will however not be considered in this thesis.
(30) a. The sky is ablaze with stars.
b. The shirt is red with blood.
3.1.1 Optional / Obligatory Arguments
Another characteristic of the swarm alternation in English regards the obligatory/optional arguments of the verb in the two different frames. According to Rowlands (2002:18), using the examples in (31)-(32) below, the optionality test shows that only the subject is obligatory in the A-subject form. The locative PP in the A-subject variant is often
habitually included, though it is not obligatory (Rowlands, 2002:18). When it comes to the L-subject form, both arguments are usually obligatory. This furthermore means that the AGENT and LOCATION in the L-subject version are arguments, whereas the
LOCATION in A-subject seems to not be an argument but a modifier.
(31) a. Bees swarmed (in the garden).
b. Fleas hopped (on the carpet).
(32) a. The garden swarmed *(with bees).
b. The carpet hopped *(with fleas).
3.2 SwarmAlt Verbs, Telicity and Holistic Effect
As some authors e.g. Levin (1993) have suggested that the swarm alternation is in a systematic relationship with spray/load, largely based on the ‘holistic effect’, I will in this section clarify the relation of SwarmAlt to spray/load. The second aim of this section is to identify two basic properties of SwarmAlt sentences, namely their atelic nature and the
‘holistic effect’. In section (3.2.2) I will establish that SwarmAlt sentences are atelic and in section (3.2.3) I will explain and discuss the 'holistic effect' in more depth.
3.2.1 The Relation of the Swarm Alternation to Spray/Load
The spray/load and swarm alternation, exemplified in (33)-(34), are often discussed in the same context. Levin (1993) for instance, based on certain aspects and/or shared features which I will present below, has suggested that SwarmAlt is in a systematic relationship with (transitive) spray/load. The first main reason is connected to the sentence patterns. Both alternations have a Locative and a with-variant which display similar patterns with the PPs. Where swarm has a location PP in the unmarked variant,
spray/load includes a LOCATION argument, and in the with-variant the parallel can be seen with regard to the with-PP, which in swarm includes what is often referred to as AGENT, whereas spray/load contains a THEME argument. Crucially though, swarm is an alternation between subject and complement position, whereas spray/load is an
alternation between VP internal arguments. According to Levin (1993:54) however, the swarm alternation can be viewed as a parallel to spray/load if the intransitive subject is seen as playing the role of the object in spray/load, i.e. Levin (1993) is proposing that the subject in swarm is an underlying object and thus, like spray/load, has two internal arguments. Levin (1993:50) thus views (intransitive) swarm and (transitive) spray/load as a single alternation by treating swarm verbs as unaccusative.
(33) a. Fred sprayed paint on the wall.
b. Fred sprayed the wall with paint.
(34) a. Bees are swarming in the garden.
b. The garden is swarming with bees.
The second reason, which according to Rowlands (2002:20) is a commonly cited main point of correspondence between spray/load and swarm, is the holistic effect associated with the with-variant in both alternations (Levin (1993:54)). In both spray/load (33b) and swarm (34b) the location is understood to be “completely” affected (Levin, 1993:50- 54). However, as will be discussed below, a closer examination shows that the holistic effect in swarm is not like that of spray/load.
3.2.2 SwarmAlt Verbs are Atelic
Contrary to e.g. spray-load verbs/VPs, which most often “imply an inherent endpoint to the event” (Rowlands, 2002:21) and are thus categorized as telic events, swarm verbs do not imply an inherent endpoint to the event and are followingly categorized as atelic events. The atelic reading of the A-subject variant where the location argument (NP2) is expressed in a locative PP refers to how the PP does not “entail movement into or out of the enclosure but only movement that takes place entirely within it” (Dowty, 2000:114) i.e. denotes some kind of activity which in theory can continue indefinitely. To illustrate, the example in (35) is mentioned by Dowty (2000:114). According to Dowty, neither of the sentences can mean that “the group of bees flew INTO the garden” (2000:114). The activity and movement of bees swarming is rather something that takes place within the location, i.e. within the garden.
(35) a. Bees swarmed in the garden.
b. The garden swarmed with bees.
That the swarm alternation is associated with and denotes atelic events, can furthermore be shown with fairly simple tests for telicity, namely the “in-for test” using durational adverbial phrases (Saeed, 2016:121). Both spray/load and the swarm alternation denote durative events, however, this test portrays different results when it comes to the telicity of the event (Rowlands, 2002:22). In order to test for telic events, one can use the adverbial phrase in x time, whereas the for x time identifies atelic events (Rowlands, 2002:22), as shown in (36)-(37).
(36) a. Bees swarmed in the garden (*in ten minutes). telic
b. The garden swarmed with bees (*in ten minutes).
(37) a. Bees swarmed in the garden for ten minutes. atelic b. The garden swarmed with bees for ten minutes.
3.2.3 The ‘Holistic Effect’ in L-subject sentences
What is referred to as the “holistic” or “affected” interpretation is a property of the alternation which refers to how there is a slight meaning difference between the two variants (Dowty, 2000:112). The L-subject form contains a sense of holistic
interpretation which is lacking in the A-subject form (Dowty (2000:113); Levin
(1993:54); Rowlands (2002:18)). Thus, whereas the A-subject form entails an event or activity located within an area, the L-subject form additionally entails that the entire location is affected (Dowty (2000:113); Rowlands (2002:18)). With regard to the example in (38) below, (38a) entails that some flies are swarming in the kitchen, whereas in (38b), the kitchen is full of swarming flies.
(38) a. Flies were swarming in the kitchen.
b. The kitchen was swarming with flies.
It has furthermore been a question of whether the holistic effect observed in SwarmAlt is the same as the holistic effect associated with spray/load, because there too, there is a slight semantic difference between the two variants (Dowty, 2000:112-113). Although this might seem to support the idea that SwarmAlt is related to the locative (spray/load) alternation (cf. Levin (1993:50)), this is according to Dowty (2000:113) only a superficial parallel. The holistic effect in spray/load is not the same as in SwarmAlt. It seems that researchers have agreed that in the case of spray/load, the holistic effect is connected to how spray/load involves telic VPs wherein the direct object measures out the event, i.e.
the direct internal argument is an Incremental Theme (Dowty (1991); Arad (2006);
Rowlands (2002:21)). In SwarmAlt on the other hand, which involves atelic verbs, the holistic effect in L-subject sentences is not because the verb is telic and is thus of different nature.
With regard to the holistic effect in the L-subject sentences in SwarmAlt, Dowty claims that L-subject sentences (e.g. The table is crawling with ants) act as characteristic descriptions of the subject NP, similar to "NP be Adj" sentences (The table is
shiny/smooth++), where the Adj predicate is normally understood as characterising the whole of the subject referent (Dowty, 2000:123-4). Both refer to surface properties of an object that “distribute to most or all sub-portions of that surface” (Dowty, 2000:123-4).
Just like the (entire) surface of a table that is smooth (or shiny) is usually smooth (or shiny), a table crawling with ants also has ants crawling in most or all parts of it. This aspect does not apply to surfaces only however, but also three-dimensional spaces (Dowty, 2000:124).
However, not all with-sentences which are regarded part of the swarm alternation exhibit this holistic effect, according to Rowlands (2002:34-35). Rowlands (2002:35) further states that in metaphorical expressions of this alternation as well as some hyperbolic readings, exemplified in (39) below, there seem to be no clear holism contrast between the unmarked and marked variant (Rowlands, 2002:19). Rowlands (2002:35) argues that whereas there is a sense of holism in the more literal sentence with the verb twinkle in (39a), there is no holism in the more metaphorical reading of twinkle in (39b).
Although I am not entering that discussion, Rowlands claims that it is because “the
referent of the NP contained within the predicate is such that it cannot be conceived of as distributed throughout the location” (Rowlands, 2002:35). Aspects regarding
metaphorical expressions of SwarmAlt will however be further elaborated in section (3.3.4) below.
(39) a. The sky twinkled with stars.
b. Her eyes twinkled with good humour.
3.3 Some Observations: Semantic Properties of the L-subject variant
In this section I will explain three significant features of L-subject sentences as characterised by Dowty (2000), and these are connected to i) the semantic classes of verbs occurring in L-subject form, ii) restrictions on the with-phrase in the L-subject variant and iii) the literal vs. metaphorical meaning distinction in the A- and L-subject variant. Furthermore, I will explain how these properties contribute to understanding SwarmAlt, focusing especially on how the L-subject sentences have a different
meaning/function from A-subject sentences (Dowty (2000), the semantic properties of the L-subject pattern and why some verbs alternate and others do not. The following subsection briefly outlines the L-subject characteristics observed by Dowty (2000), before discussing them more thoroughly.
3.3.1 Dowty’s (2000) L-subject Observations: “Five Characteristics”
Dowty’s (2000) analysis constitutes five characteristics or “general observations”
regarding certain semantic restrictions on the L-subject variant, which he claims represent “factors that distinguish acceptable from unacceptable cases” of the swarm alternation (Dowty, 2000:114). From Dowty’s perspective, these general observations are important characteristics of the swarm alternation in English, and in order to
investigate similar aspects in the Norwegian counterpart, the following sections will focus on and outline the relevant characteristics.
What distinguishes the L-subject form from the A-subject form according to Dowty (2000:120) is that:
(i) it occurs with a semantically well-defined class of atelic verbs which denote
perceptually simple activities usually recognizable from temporally and spatially limited input
(ii) the denotation of its with-phrase object must be plural and indefinite (in understood interpretation if not in syntactic form)
(iii) at least for sound verbs, the with-phrase object must refer to perceptions (sound) themselves, not agents that produce them (just the opposite of the A-subject form) (iv-v) the L-subject form is more suited to metaphorical use than the A-subject form. To this the familiar traditional observation about the L-subject form can be added: (v) it entails that the activity fills the space denoted by the Location subject, whereas the A- subject form does not.
These observations illustrate that there are some very specific semantic restrictions on the L-subject variant, which, according to Dowty (2000:120) implies “beyond reasonable doubt […] that the L-subject form somehow has a different semantic function from the A- subject form”.
3.3.2 Dowty’s (2000) Five Meaning Classes
According to Rowlands (2002:27), although the substantial research on the swarm alternation is not very extensive, there is much consensus when it comes to the verb classes associated with the alternation. Verbs associated with the swarm alternation are listed in Salkoff (1983), Levin (1993) and Dowty (2000), however as Dowty’s (2000) classification of verb classes associated with SwarmAlt in English is derived from both Levin and Salkoff (Rowlands, 2002:28), I will here mostly focus on the semantic classes outlined by Dowty (2000).
Whereas Levin (1993) lists seven semantic classes of verbs associated with the swarm alternation, Dowty (2000) has reduced to five semantic classes. These classes include (i)
‘small scale’, usually repetitive, visually recognizable physical movements; (ii) animal and other perceptually simple sounds; (iii) conceptually simple visual perception of some kind of light emission; (iv) smells and (v) predicates indicating degree of occupancy or abundance (Dowty, 2000:115).
Dowty (2000) goes on to further elaborate on each semantic class, however, what is clear from his elaborations as well as the list above in more general terms is that for at least four of the semantic classes listed, they all share the common characteristic that they denote “perceptually simple activities usually recognizable from temporally and spatially limited input” (Dowty, 2000:116). In other words, they denote (literal as well as metaphorical) activities/events that are easily and/or immediately perceptible to our senses, they are either seen (i, iii), heard (ii) or smelt (iv). For instance, noticing whether a fountain is foaming (40) is something that happens in an instant, as opposed to the slightly more extended time and cognitive analysis it normally takes to determine that e.g. a cow is grazing (Dowty, 2000: 115). A verb from each class is illustrated in (40)- (44) below.
(40) a. Soap is foaming in the fountain. b. The fountain is foaming with soap.
(41) a. Flies are buzzing in the window. b. The window is buzzing with flies.
(42) a. Diamonds are glittering on her b. Her dress is glittering with diamonds.
dress.
(43) a. Alcohol reeked on his breath. b. His breath reeked with alcohol.
(44) a. Fish are teeming in the pond. b. The pond is teeming with fish.
Dowty (2000:116) furthermore illustrates that verbs that are similar to those included in the list of perceptually simple activities but denote more ‘complex’ actions, e.g. such as resound vs. harmonize in (45), often sound odd, bizarre or are not acceptable in the L- subject form. In addition to almost exclusively falling into these five semantic classes, the verbs appearing in the L-subject form are all one-place (atelic) process/activity verbs (Dowty, 2000:116).
(45) a. The voices of the choristers resounded in the church.
b. The church resounded with the voices of the choristers.
c. The voices of the choristers harmonized in the church.
d. *The church harmonized with the voices of the choristers.
According to Rowlands, as well as Dowty, there are however some peculiarities associated with some of the verbs within the identified classes (Rowlands (2002:32);
Dowty (2000:118)). One of these peculiarities is connected to sound verbs and their
with-phrase NP. Some of the sound verbs requires and are more acceptable with a simple NP denoting the sound itself rather than the producer of the sound (46), whereas others are more acceptable when the with-phrase NP denotes the producer rather than the sound (47) (Rowlands, 2002:32).
(46) a. ?The hall twanged with guitars.
b. The hall twanged with the sound of guitars.
(47) a. The eaves twitted with sparrows.
b. ?The eaves twitted with the chirping of sparrows.
Other peculiarities are associated with verbs of entity-specific modes of being, which are verbs such as blossom, flower and bloom (Rowlands, 2002:33). With a simple NP in the with-phrase, especially one which is unrelated to the verb, these sentences sound odd (48), however, they are acceptable if the NP is modified (49) (Rowlands, 2002:33).
These verbs are moreover, according to Rowlands (2002:33), the only group of verbs among the semantic classes identified which systematically allows the with-phrase to be omitted (50).
(48) a. ?The bush bloomed with blooms.
b. ?The garden flowered with flowers.
(49) a. The bush bloomed with thousands of fluffy white blooms.
b. The garden flowered with all kinds of colourful flowers.
(50) a. The bush bloomed.
b. The cherry tree blossomed.
3.3.3 Restrictions on the With-phrase in the Marked Variant
In focus in this subsection is Dowty’s (2000) observation regarding specific restrictions on the with-phrase in the L-subject variant. Dowty notes the requirement, which the unmarked variant does not have, that the object of with in acceptable L-subject sentences must be a bare plural or mass term (Dowty, 2000:117). A definite or
quantified NP, including quantified indefinite, is thus not possible. A quantified indefinite is one which specifies a particular quantity or amount, including a singular countable noun with an indefinite article (a bee) or a numeral (two bees, a thousand bees) or other more vague quantifiers like many and several. The with-phrase restrictions are illustrated in (51)-(52).
(51) a. Roaches crawled on the wall. Unmarked variant b. A roach crawled on the wall.
c. The wall crawled with roaches. Marked variant d. *The wall crawled with (a roach/a thousand
roaches/the roaches/those roaches).
(52) a. Soap was foaming in the bathtub.
b. The bathtub was foaming with soap.
c. *The bathtub was foaming with (the soap/a lot of soap).
However, Dowty (2000:117) also illustrates that there are acceptable instances where the object of the with-phrase is a singular NP, such as in (53) below. The reason why such cases are acceptable according to Dowty (2000:117) is that the sentence involves
